With the development of semiconductor technologies and standardization of wireless local area networks (WLAN, Wireless Local Area Networks) in 802.11 standards, costs of WLAN technologies are greatly reduced, and the WLAN technologies are also increasingly widely used. At present, versions of 802.11 standards have evolved from 802.11a/b to 802.11g, 802.11n, and 802.11ac, and an STF of a preamble in 802.11 ac performs initial time synchronization estimation. To ensure backward-compatibility and interoperability between products of different versions of 802.11 standards, a mixed format (MF, Mixed Format) preamble (which is referred to as preamble for short below) is defined from 802.11n. A legacy part of the preamble includes a legacy field that is the same as that of 802.11a.
FIG. 1 shows a structure of an existing preamble. As shown in FIG. 1, the legacy part of the preamble includes three fields, that is, a legacy-short training field (L-STF, Legacy-Short Training Field), a legacy-long training field (L-LTF, Legacy-Long Training Field), and a legacy-signaling (L-SIGs Legacy-Signaling) field, where the L-STF field is used for frame-start detection, automatic gain control (AGC, Automatic Gain Control) setting, initial frequency offset estimation, and initial time synchronization; the L-LTF is used for more accurate frequency offset estimation and time synchronization and is also used to perform channel estimation to demodulate the L-SIG more accurately; the L-SIG field is mainly used to carry data rate information and data length information, so that a receive-end device can determine, according to the data rate information and the data length information, a length of data carried in a same frame as that of the preamble, so as to determine a proper time for staying idle.
A short training sequence carried in the L-STF field from a transmit end includes multiple sub-sequences of a repetitive structure, and therefore has relatively strong autocorrelation. Therefore, at a receive end, initial time synchronization can be performed by using a falling edge of an autocorrelation value obtained through autocorrelation processing (in other words, an autocorrelation operation) performed on a received signal (which includes the short training sequence).
Besides, to increase robustness of an algorithm, in addition to the autocorrelation processing mentioned above, cross-correlation processing (in other words, an autocorrelation operation) may also be performed on a received signal and a locally known short training sequence. Because the short training sequence has optimal autocorrelation property, multiple peak values may be obtained through cross-correlation processing. Besides, as described above, a falling edge may be obtained through autocorrelation processing, and therefore an initial location of the L-STF field can be determined with reference to the last peak value and the falling edge.
However, a calculation amount of the foregoing processing is relatively large, calculation complexity is relatively high, and fast initial time synchronization cannot be implemented.